The Trump administration aims to enhance the efficiency of the US government by leveraging artificial intelligence (AI) tools. With the rise of large language models (LLMs) like ChatGPT, understanding the implications of AI in various sectors has become increasingly critical.
As AI continues to permeate everyday life, questions surrounding its appropriate application emerge. Should students rely on AI for academic writing? Could it serve as a replacement for therapists? And importantly, how can it enhance governmental efficiency?
In both the US and the UK, these inquiries are gaining traction. Under the Trump administration, a new initiative led by a government efficiency task force is streamlining operations, which includes the implementation of a chatbot named GSAi aimed at supporting federal employees. Concurrently, the UK Prime Minister has heralded AI as a transformative tool that could significantly benefit the state.
While there is potential for automation to improve government processes, the efficacy of LLMs as a solution raises more questions than answers. Recent developments, such as the release of AI interactions from a UK government secretary, highlight ongoing debates about data privacy and the role of AI in policymaking. Surprisingly, these interactions were deemed subject to freedom of information laws, similar to communications between ministers and civil servants, like emails or messaging applications.
The release of these records implies a belief within the government that engaging with AI can be akin to human interaction. However, the lack of substantive reliance on LLMs for serious policy formulation raises concerns. Notably, AI scientists caution that current LLMs lack genuine intelligence and may produce misleading information, reflecting biases in their training data.
Research among AI experts indicates skepticism regarding the path to achieving artificial general intelligence (AGI), with a significant majority doubting the feasibility of existing methodologies. This necessitates a reassessment of how we view LLMs—rather than seeing them as intelligent entities, they should be regarded as innovative cultural technologies that aid humans in processing accumulated knowledge.
Considering this perspective, the application of LLMs in government operations appears promising, provided they are managed by individuals aware of their capabilities and limitations. Furthermore, the question of whether chatbot interactions should fall under freedom of information regulations remains complex, requiring careful consideration of existing protections for ministerial discussions. Lastly, the inquiry into whether machines can think, as originally posed by Turing, remains unanswered—at least for now.
Topics:
